Assessment Of Marine Phytoplankton Using In Vivo Excitation-emission Matrix Fluorescence Spectra

In this study the feasibility of using in vivo excitation-emission matrix (EEM) spectra of living phytoplankton for determining abundance of specific classes of phytoplankton was investigated. The EEM spectrum method, with a high sensitivity, can give total fingerprints information in the whole range of wavelengths. It is a rapid analysis method that can provide qualitative and quantitative information. On the basis of the former research work of our group, this paper aims to explore spectrographical classifications of phytoplankton, which is at the level between division and genus in Biology, and set up a rapid fluorescence analysis method to identify the most abundant phytoplankton class and assess its abundance.Twenty-five phytoplankton species, representing five divisions in the East China Sea were chosen, including seven Dinophyta species: Alexandrium tamarense, Prococentrum marinum, Prorocentrum micans, Gymnodinium sp., Gymnodinium simplex, Prorocentrum minimum, Scrippsiella trochoidea; ten Bacillariophyta species: Pseudo-nitzschia pungen (PS0201-01), Skeletonema costatuma, Coscinodiscus sp., Cylindrotheca closterium, Odontella cf_sinensis, Chaetoceros curvisetus, Chaetoceros debilis, Chaetoceros didymus, Thalassiosi rarotula and Ditylum brightwellii; two Chlorophyta species: Chlorella pynenoidosa and Platymonas helgolanidica; two Cyanophyta species: Synechocoocus sp. and Anabaena sp.; four Chromophyta species: Phaeocystis globosa, Rhodomonas sp., Rhodomonas salina and Chattonella marina. The species were cultivated in lab under different temperatures (25℃, 20℃, 15℃) and illumination intensities (15000 lux, 10000 lux, 7000 lux, 4100 lux, 1100 lux). The main research work is as follow:1. The classifications of marine phytoplankton using EEM fluorescence spectra were studied. On the basis of chemometrics methods obtained from the former research work, after the Rayleigh and Raman scattering peaks were removed from EEM, characteristic spectra were obtained by using principle component analysis (PCA). Two indexes mean correlation coefficients R and the first component contribution to explore the spectra similarity. The results indicate that fluctuating range of was larger than R. In order to classify phytoplankton, first, one or more representative spectra were obtained for each species according to the similarity measurements and hierarchical cluster analysis (HCA) of characteristic spectra, and 30 representative spectra were obtained in total. Then the twenty-five phytoplankton species were spectrographically sorted into different"spectral classes"by means of HCA, projection discrimination based on PCA applied to the representative spectra and Bayesian discriminate analysis, ratio analysis to the characteristic spectra. The definition class is between division and genus in Biology.2. The feasibility of utilizing standard Chl a to explore the method detection limit was studied. Firstly, the mean of fluorescence intensity of the emission wavelength 675 nm, the excitation wavelength 380, 410 and 430 nm were selected to represent the fluorescence intensity of the corresponding Chl a sample. Thereby, the standard curve was obtained. The detection limit was 0.065μg/L by IUPAC, and the qualitative and quantitative detection limit was 1.85and 2.77μg/L respectively by ISO criterion. Secondly, the method detection limits was explored by using standard Chl a and in vivo EEM spectra of phytoplankton. The mean of fluorescence intensity of emission wavelength 675 and 680 nm, excitation wavelength 440 nm were selected as the fluorescence intensity of sample and corresponding with the fluorescence intensity of the standard Chl a. The phytoplankton biomass was expressed with concentration of standard Chl a. The first is the theoretical detection limit, which is according to the IUPAC criterion. The detection limit was 0.10μg/L by IUPAC. The practical detection limit for nine selected species was between 2.5-20.0μg/L.3. The identifications of 660 single species and 105 phytoplankton mixtures were studied by NNLS. With the threshold value 0.2, the correct recognition rate of single species was nearly 100%. The correct recognition rate was 87% for all mixtures. Picking out the samples whose concentration were less than the practical detection limit, the qualitative correct recognition for rate 95 mixtures left could be up to 95%.This study suggests that the EEM spectral technique is a rapid efficient tool for rough phytoplankton assessments.